What value does this provide over having a workflow like:
loop(implement -> review) with sub agents
I'm not seeing why I would ever want to add another tool like this. The very concept seems like a PITA to get working with the isolation / sandboxing, and likely too much security risk / complexity.
loop(implement -> review) with sub agents
I'm not seeing why I would ever want to add another tool like this. The very concept seems like a PITA to get working with the isolation / sandboxing, and likely too much security risk / complexity.